Any updates on the XPS 16 heating/ throttling issue?

The test doesn't really show that it's power consumption or heat throttling. The screen was off and the system was being cooled, which is both variable suspects being killed at once.

By comparison, RGBLED on with 4670 is probably 30 or more watts.

 

I do think one of the issues is heat, but it doesn't seem to be CPU or GPU heat directly. It sometimes throttles the CPU for me when the GPU/CPU temps are 98/86C and in another game 95/80C. I notice different programs read different temps (around 3C difference). Is there a particular program that is "most accurate" or are they reading different sensors?

I notice that the middle of the bottom of my laptop gets insanely hot (literally hurts to touch for more than a second). I'm not a wimp either when it comes to heat- by comparison the power adapter feels luke warm when at full load.

 

I think HWmonitor by CPUID is the most accurate. Gpig, your temps seem really high and I thought mine was too hot to touch without getting burnt (it is) @ 84 C CPU and 84.5 C GPU. I have been playing games on an external monitor all day long without ANY throttling (laptop screen off and vent not being blocked with back raised) and my max temps have been under 75 C. I used to get throttling on Bad Company 2 in this scenario but today I have played lots of single player and multiplayer and no throttling. I will try and make a throttlestop log and post it...this is freaking RANDOM.

 

One problem is every BIOS revision is potentially different. A10 is pretty much unusable for me because it hits the GPU max temp very fast (I hit 86C but other people say it's less). And then there are different parts (graphics card, screen, CPU). There are so many configurations and questions. Here's one: why does no one with an i7 820QM report CPU throttling? Are those immune?

 

820QM users do have throttling. Their default multiplier is 13 and you'll find their logs dipping into low 12s during throttling.

 

That's where the CPU heatsink is. Since the laptop is so thin, you're only an inch or two away from something that is 70C. Probably would feel hot.

~Ibrahim~

 

just got my motherboard replaced still throttling on the gpu. They want to replace the motherboard again.

 

Considering Dell don't like Furmark+Prime, I was playing around trying different programs and noticed that BOINC uses 8 threads. Installed it and started monitoring using TS: it shows 99.9% utilisation and the Multi at 12.8 (fairly stable). It won't reach 13 unless you introduce some chip/clock modulation (thanks UW).

So this is less intense than Prime and is a real-world application. It was fairly late so I just started a game to see how they play together and the Multi dropped to 11-12.

The game alone gives me Multi of 16 but if I run it while I have BOINC running, it's between 11-12. (as a reference: Prime+Furmark is between 10-11.5). If someone can expand on that, that could also be used to show Dell this system throttles with real-world applications (it's not meant to go below 12!).

add: should have mentioned that the CPU tries to keep running in Turbo (55.04W) with BOINC alone but keeps dropping to 45W (so one assumes that's why it keep cycling down at 12.8 Multi instead of 13). Wonder what program actually allows full stable Turbo with 8 threads running at almost max.

 

I don't know if your 45W and 55W numbers are 100% accurate. It sounds like an approximation which might not be accurate when the CPU starts rapidly cycling turbo boost on and off.

The lack of turbo boost when all 8 threads are loaded is probably an Intel limitation and not Dell throttling. Turbo boost is only available when CPU power consumption is within the Intel spec.

I wrote a testing program a while ago that lets you fully load your CPU but it does it very efficiently so it won't create as much heat or use as much power as most testing programs do. It's called LoadTester and is hiding in the RealTemp download.

http://www.fileden.com/files/2008/3/3/1794507/RealTempBeta.zip

LoadTester can be used to create a variable load from 0% to 100% in steps of 10% but only lets you load one thread. To fully load all 8 threads, you need to run 8 instances of LoadTester. It's a handy tool for testing how turbo boost works and if you want to, you can also use Set Affinity... to lock each instance to a different core. You should be able to fully load your CPU with this program and still get full turbo boost which would be a steady 13.00 multiplier on an i7-720.

 

Thanks, well I gave that a go and TS shows it's using 99.9 of each thread. I then set affinity (different for each of the 8 entries) and it remained the same (not 100%).

It's showing Multi of 12.2 for me and Hardware Monitor shows that same cycling for Watts - between 45 and 55.04.

Whatever this power level limit is, your program uses less than Prime (which is stable 12 and 45W all the time) but more than BOINC (which also showed 99.9% utilisation but with Multi around 12.8).

 

Thanks for testing that. It sounds like you are just hitting the TDP turbo boost limit of the CPU. Here's what the Intel Turbo White Paper says about turbo boost:

1.2 Dependencies / Algorithm

Intel® Turbo Boost technology core frequency upside availability is ultimately constrained by power delivery limits, but within those constraints, it is limited by the following factors:

• The estimated current consumption of the processor
• The estimated power consumption of the processor
• The temperature of the processor

 

no problems, thank you for giving us all these various tools!

So I'm still trying to find a program which gives me 13 Multi on 8 threads, just tried Fritz Chess Benchmark and that's the same: Multi between 12.4-12.6 and Watts changing between 45-55.04 (utilisation at 99.9%).

TS version 2.54. Anyone have any programs where they can get Multi of 13?

 

Resident Evil 5 benchmark demo gives me steady 13x (it's a free game demo that runs itself).

 

Thanks, I gave that a go. The DX9 test results are much worse than DX10, is that normal (fps point of view)? Would have thought it was the other way around, but anyway:

It doesn't utilise the 8 threads all the time, so while there were some areas showing 13 Multi, it kept jumping up and down (like a normal game).

Trying to find at least one program which proves that this system can run Multi of 13 with 8 threads at 99.9% or 100%.

 

Hi guys,

I just got my new XPS 16 and I want to know whether my CPU GPU temp are fine.

But I could not find any XPS Thermal Monitor software. Tried to find from external web like filehippo, I could not find any thermal monitoring software (free) as well.

Can someone help to suggest me what software I can use?
I googled and found that XPS Thermal Monitor is no longer supported by Win 7 64bit.

 

I have this problem, but thought it was because it calculates the average multi.
My multi is all over the place even with multi 13 set and TS turned on

 

That's funny, in the 3 computers including my XPS 1645 I've run the benchmark demo on, the DX9 test ran with about 20% more fps (on both of the in-game tests with default settings).

While it may not be using/loading many threads, I found steady multipliers of 13x (but you're right that the CPU utilization is no where even close to 100%). I thought the non-13x multiplier sections were during load sequences but I'm not sure now (it loads at least 3 times during the 'variable test'). Here's my (old) log:

 

On a Core i7-720QM, when you set ThrottleStop to 13, you are asking the CPU to use the default 12 multiplier and the extra + 1 tells the CPU to use as much turbo boost as possible. When 3 or 4 cores are active, you will get a maximum multiplier of 13 as long as CPU power consumption is not over the turbo boost limit. When you go over the CPU power limit, turbo boost gets turned off and the multiplier drops back to 12. When less than 3 cores are active, more turbo boost is used so the average multiplier will go higher than 13. You need just the right load and CPU power consumption to get a consistent 13.00 multiplier.

 

Even at idle it doesn't stay at 12 or 13 it's more 8x or 9x..

 

At idle, cores are constantly going into and out of sleep mode so the average multiplier for each thread can drop to as low as 7 on a Core i7-720QM.

If you want a low and steady multiplier at idle then try using the Power Saver option. That might help.

The multiplier in a Core i CPU is very dynamic and can be changing a hundred times a second. Programs that show a nice steady multiplier are rounding things off or are not being 100% honest.

 

Then what's the effect of turning on 13 multi in TS? It doesnt change anything here..

 

Well, if you set the multiplier to 12 then you are telling your CPU not to use any turbo boost. That will definitely make a big difference at full load and it also makes a difference at idle. Your idle screen shot shows that one of your threads is using turbo boost because it is showing an average multiplier of 13.61.

 

Okay...HELP!!!!

Just got my replacement 1645 last night with the 5730 GPU, and straightaway, trouble.

I have no idea how this hasn't been brought up till now, but the damn GPU throttles like the axe. And it freaking throttles at 78 deg C. With ANY BIOS! Even A09. I know that A09 updates the video BIOS for the 4670, but DOES NOT for the 5730. Even A10 has me stuck as some GPU BIOS from 20/11/2009 (or is it just me? By some freak chance?)

Which leaves my GPU downclocking to 450MHz as soon as I hit 78 deg C, and even worse is, it does not rise back up when the temperatures fall.

I've been trying to get Rivatuner to work so that I can lock the clocks, but haven't had any success so far.

Hows it going for the rest of you 5730 owners? Any way you've gotten Rivatuner working? Or has anyone else seen a video BIOS update?

PS - Another weird thing, I noticed that I'm seeing some throttling with the slimline PA4E 130W adapter I got. But the PA-13 brick I had with my old SXPS16 hardly causes throttling. I'm sticking with the brick for now I think

 

I get the same thing regarding the 78* thermal limit... PRETTY DANG GAY OF YOU DELL!!!!!!!!!

 

@ all 5730 owners

Temporary fix found for people who are throttling at 78 deg C ONLY FOR HD 5730. Clocks can be locked using Rivatuner. This is how:
1. Download Rivatuner
2. Download Atsiv and Rivatuner batch file from the following source
Atsiv.zip - download now for free. File sharing. Software file sharing. Free file hosting. File upload. FileFactory.com
3. Extract zip to Rivatuner folder.
4. Open Rivatuner.cfg using Notepad
5. Scroll down to RV770 in the [GPU_1002] section, add a comma and the string "68C0h".
6. Start Rivatuner using the Rivatuner.bat file that was extracted earlier.
7. You should have Rivatuner running now with 128bit, RV770 (400sp) below your GPU name. Click on customise, then Low lever system settings (leftmost box)
8. Click on checkbox Low level overclocking>Detect now>set clock speed to 651 (yes, 651, not 650).
9. APPLY! It will NOT downclock at 78 deg C now

This hack works despite the fact that 5000 series cards are not yet supported by Rivatuner. Not very elegant, but atleast you can freakin use that 5730 now

 

Props to you man! REPPED!!!

 

Has anyone looked at Motherboard temps (i.e. TZ01 in HWMONITOR) as one cause of CPU Throttling? It seems as though my system may throttle the CPU when TZ01 hits ~84C on BIOS A09. I wouldn't be surprised if Dell accidentally (or purposely) moved the 84C limit on the GPU in BIOS A08 to the Motherboard and instead decided to throttle the CPU. This 84C value seemed to sync up in the 2 games I tried as well as Prime+Furmark (min brightness & stuff to avoid instant throttling due to power).

I'm not sure that the Motherboard temp is used to throttle the CPU, but the next time someone (particularly XPS 1645/BIOS A09/4670 people) is playing a game and experiencing CPU throttling, take a look at TZ01.

 

It does seem when I do throttle (ONLY when playing certain games on the laptop screen/vent blocked) that the highest I've seen the temp on the CPU go to is 84 C with the GPU being 84.5 C. The GPU seems to throttle randomly under BIOS A10 but either it's also only in certain games or happens only sometimes (maybe its just the easier games to run that I'm not noticing a significant frame-rate drop?). I am at least happy I can play any game now on the external monitor without throttling, except for GTAIV which runs like crap reguardless...

 

@Wetcardboard

You say your GPU hits 84.5 deg C max. Can you confirm that your GPU is throttling when temperature goes over 78 deg C? Or are you running at normal speeds at 84.5 deg C? Because, for many 5730 users, the GPU throttles at 78 deg C and temperature does not rise above that.

 

Is anyone else using Riva Tuner and having the system shutdown????

I was running a stress test prime+furmark while using RivaTuner as suggested and all seemed great after I got above 78*... but not once but TWICE my system shutdown when it the GPU in furmark read 88-89.

Well, anyone else shutting down @ 88-89C?? This sucks...

EDIT: Oh, Mr Hohoha the GPU was NOT throttling above 78.

 

@cbaty

Whats your ambient temperature? In a room at 25-26 deg C, I only could hit 83-84 deg C in Furmark+Prime95 after 25 minutes of testing, and Crysis only got me to 80 deg C after half an hour!

My old laptop with the 4670 used to shutdown at 93 deg C. Maybe they've lowered that limit on the 5730, also considering the fact that it runs a lot cooler than the 4670

My advice would be to test real world performance rather than synthetic benchmarks. Does it go that high while gaming? (I normally use Crysis for testing, everything except shaders set to high - it stresses both CPU and GPU properly). Also, while playing Crysis, I had to activate Throttlestop as the laptop starts to see CPU throttling when GPU hits 79-80 deg C.

Sad to see that the 1645 w/ 5730 continues to have the same issues that plagued the old 1645s.

 

How is it a synthetic benchmark, it IS a benchmark! haha

The room temp is about 24-25C.

Real world throttling happens in bad company 2 very noticeably.

With my 4670 I would get to 98C and so far the 5730 can only get to high 80's.... Well without the system just completely freaking out and turning off....

 

The laptop shouldn't just power down because of light heat on the GPU. Those temps you mention are no where near the maximum operating temperature of the card. This is not normal.

My 4670 throttles at 100C (Furmark), but never just powers down the laptop.

Just guessing, but forcing the clock speed in Rivatuner could possibly confuse the motherboard on how much power it needs, causing a crash (when it hits the Dell thermal limit of ~78, maybe it thinks it needs less power)?

 

Yeah I know it is not supposed to do that haha, I just don't have a clue as to what is happening :/

 

@cbaty

Shutdown temperature varies from computer to computer, and also by ambient temperature. I have a hunch that it basically has to do with how much heat is transferred to the rest of the motherboard from the GPU (TZ01 in HWMonitor). Which is why mine used to shut down at 93 deg C, even when other people could push theirs till 100 deg C.

You could have Dell change your thermal pads on the GPU. That helped my old 4670 run cooler.

 

I don't want to be rude, but how can ambient temperature change the thermal limit where the laptop powers itself down? We all know the ambient temperature has a great affect on temperatures of the CPU, GPU and laptop in general, but how would this change the maximum allowed temperature? Seems to me a higher room temperature just makes the laptop more likely to reach the set thermal limit.

 

@gpig

Don't think you quite understood what I wanted to say. This is how heat is supposed to flow:

GPU>Heatsink>Pumped out of case by air

However, heat also flows from GPU>Northbridge+Southbridge>Case>Surroundings

If ambient temperature is higher, case temp is higher. That means, by principles of Newton's Laws of Cooling, the rate of heat dissipation from mainboard to case is lower. So for X GPU temperature, the ambient temperature will affect how hot the rest of the motherboard gets. If ambient temperature is high, for the same GPU temp, mainboard temperature will be higher. Which is why, for some people, the critical MOTHERBOARD temperature will be achieved earlier than others. What I'm trying to say is that there probably exists a fixed max safe operating temperature for the mainboard, which being crossed is what causes the system to switch off.

I don't pretend to be an expert on cooling, and like I already said, this theory is no more than an educated hypothesis. You tell me if you have a better explanation.

 

Alright, your explanation sounds reasonable.

If you want me to take a guess, I'd say the GPU was defective when it left ATI's factory due to whatever reason.

 

Well I don't think it has to do with the ambient room temp at all.. the last 1645 got to almost 100...... I would agree that ambient room temp would help the temp rise faster, but in no way should it affect at what temp exactly the laptop shuts down really... I think the components inside make the laptop hotter than the room in no time at all before it even starts to throttle...

Also I have already pulled the screens and the thermal pads and placed ICD7 on there.... Idle temps are great at I have no real reason to believe the GPU is defective.. why do you think so gpig?

 

I don't think your GPU is defective. I think mrhohoha's 4670 that powered down the laptop when it hit 93C is defective.

 

Oh I see.

Well I can confirm that when my GPU hits 89c the whole system just shuts down. The system even had to do some restore task.. not a full system restore but restored something...?

I did this running only furmark!!!

 

For me at least, the TZ01 follows the CPU temps very closely and most of the time matching them - it doesn't follow GPU temps.

You can easily check this by simply running Prime to warm up the CPU while GPU remains as-is, you should see TZ01 nearly matching your CPU temps. Or run furmark only and TZ01 doesn't rise to match GPU.

As a side note, can anyone see any difference between Passive and Active cooling Policy? Mine behaves exactly the same way (Multi, temps, fan, Watts, etc) regardless of this setting.

 

@gpig

Well, I would have thought my GPU/mainboard was defective too. But I had three motherboard replacements, and my old SXPS16 shut down right after it hit 93 deg C on all three motherboards, ie. three different GPUs (judging from Rivatuner logs). This however, was when my ambient temperature used to be 32-35 deg C.

When I moved to an airconditioned room, I saw it going to 95 deg C, which is when I closed my game manually. (Didn't want to see meltdown)


@cbaty
Well, like I said already, ambient temperature DEFINITELY affects how hot your laptop runs on an average, and especially the SXPS 16 (which has a magnesium alloy casing). You'd be surprised to hear that about 25% of the heat dissipated is from the metal casing itself. If the surroundings are warm, the rate of heat dissipation automatically decreases (like I said, basic thermodynamics), which leaves more heat trapped inside the case, which causes the temperature of the mainboard itself to rise above set limits.

I've seen this in practice, and if you don't believe me, try for yourself - run it in a hot room and you'll see.

On a more practical note, if even ICH7 does not fix the amount of heat you're seeing, your GPU is probably faulty. The max I've seen my 5730 go so far is 83 deg C (on stock cooling). I say you should call Dell for a new motherboard

PS- Um, could it be that the ICH7 you've applied maybe didn't spread properly? How were your temperatures with stock cooling?

 

Hey man, I never said it couldn't affects the temps of the system AT ALL... I am just trying to point out it shouldn't change the thermal limits on the inside...

OK, I can understand something happening a FEW degrees before usual due to it being hot out.. but these parts are getting way warmer than any air we are next to... leaving me to believe that when we are stress testing the system could care less about ambient room temp and cares more about core temps... right?

No, the compound is on good, I am sure.. I have done it many times and my idle's are hight 40's....

 

Hello all, I am a long time reader and a first time poster, just got my XPS 1645 a few hours ago and I just ran Prime95 and Furmark (I don't know if I did it right though) but anyways my specs are
i7 720
8 gigs of RAM
ATI 5730
and the 500 GB hard drive at 7,200
BIOS A09
130 W adapter
Ambient Temp was around 74 F
GPU max was 79 (I've tried to stay up on this as much as I can, but I just don't understand why my multiplier goes from 7.0 in the middle of the test, back to 12.0 at the end, the GPU hit 79 C around the time that my CPU started throttling, then when the GPU went back to 76 C my mult went back to 12.0. Can the GPU hitting 79 C cause the CPU to throttle?)
I didn't do a clean install or anything but here is my TS log
View attachment ThrottleStopLog.txt

 

Throttling can be based on GPU temperature. It's likely that it throttles for a while and then once the GPU temperature has cooled down, throttling goes away and the multiplier returns to normal. It can cycle over and over like this.

A 7.0 multiplier combined with clock modulation at 12.5% has your CPU running at the equivalent of about 115 MHz. That's not too good and definitely not what you paid for.

You can try adding GPU=2 to the ThrottleStop.ini configuration file so it logs your ATI GPU temperatures as well. It might be easier to see when or if the GPU temperature is triggering throttling.

Better yet, why not contact Dell and tell them they can have their throttling laptop until it is redesigned so it doesn't throttle.

 

79 C max gpu temp well running P95+Furmark is insane...my temps are ten degrees higher than that under stress testing.

 

@Unclewebb
Thanks for the advice Unclewebb, I will most certainly include that in my next log, I understand that 115 MHz processor isn't what I paid for, the weird thing is the fact that I saw no drop in FPS or any other indication that my computer was throttling other than viewing the log file, I also began to question the temp on my CPU might be the cause as well, is there any way to incorporate that into the log as well?

@Wetcardboard
I was shocked myself, after reading about your XPS experience, I did some testing on my desktop with a 8800 GTS, and an AMD dual core 3.0 GHz processor, the name escapes me right now, and I was hitting temps of 91 C so 79 was quite the suprise

 

The DTS data coming from an Intel CPU is like a reverse thermometer. As the CPU gets hotter, the data coming from the digital thermal sensor (DTS) counts down towards zero. This sensor is designed to trigger thermal throttling and thermal shutdown. For an i7-720QM, this sensor will read zero when the core temperature is at 100C. Thermal shutdown is designed to happen about 25C beyond this temperature.

To convert DTS data to core temperature data all you have to do is go into the ThrottleStop.ini configuration file and enter:

TJMax=100

By default, ThrottleStop reports the sensor data as is just so it is easy to see that you are still 20 degrees away from the Intel designed thermal throttling point but if you prefer to see this data displayed and logged as core temperatures, that's not a problem.

 

ThrottleStop works with i7 720qm? the last one i tried said i had to set a minimum to 100 and i didnt want to try it. TS is used to increase the amount of heat the cpu can take before slowing down right? are there other stuff it can do?